OpenAI just announced something that changes the math for law firms: Presence, a service where their engineers build and maintain AI agents for your business. You don’t hire data scientists. You don’t train models. You describe the work you want automated, and OpenAI embeds a technical team to make it happen.

For law firms, this matters because the two biggest barriers to deploying AI agents have been technical complexity and ongoing maintenance. Most partners I talk to know exactly which tasks bleed time and money: after-hours intake calls that go to voicemail, junior associates spending three days on first-pass document review, matter triage sitting in someone’s inbox for 48 hours. The problem was never identifying the work. It was building the system and keeping it running.

Presence removes that barrier. It’s a managed service model applied to enterprise AI. You get voice agents that answer every client call, ops agents that route and summarise incoming matters, and document-review agents that produce associate-grade memos. OpenAI’s engineers configure the system to your practice, integrate it with your case management and calendar tools, and stay on to tune and troubleshoot.

This is the first time a major AI lab has offered this level of hands-on service to mid-market businesses. It’s not a SaaS product you sign up for and figure out yourself. It’s closer to hiring an AI team without the payroll, the ramp time, or the risk that they leave six months in.

Let’s walk through what this looks like for a law firm, the specific work it targets, and how you evaluate whether it makes sense for your practice.

The Work That Leaks Revenue Every Week

Most law firms doing $1M to $25M in revenue lose between $80K and $250K a year to unbilled administrative time. That’s not a made-up number. It’s the sum of hours spent on intake, matter triage, document prep, and follow-up that never makes it onto an invoice.

The pattern is consistent across practices. Partners and senior associates spend four to six hours a week on work that can’t be billed: answering intake calls during lunch, reviewing conflict checks, triaging emails from potential clients, summarising documents for a junior associate to review later. Junior associates spend even more time on first-pass document review, contract analysis, and discovery prep. That work is billable in theory, but clients push back on the rate, or the firm eats it to stay competitive.

After-hours intake is the clearest example. A high-intent caller phones your office at 6:30pm on a Tuesday. They’ve been injured, they’re facing a commercial dispute, or they need estate planning before a deadline. Your voicemail picks up. They leave a message. By the time someone calls back the next morning, they’ve already spoken to two other firms. Conversion rates on after-hours intake sit around 30% at best. The other 70% are gone.

Document review is the second major leak. A junior associate bills at $200 to $400 an hour, depending on market and seniority. First-pass review of a contract or a discovery batch takes six to twelve hours. The client sees the invoice and questions the value. The firm either defends the rate and risks the relationship, or writes off half the time. Either way, the economics don’t work.

Matter triage is less visible but just as expensive. Intake forms and emails sit in a shared inbox. Someone has to read each one, figure out the practice area, check for conflicts, decide which partner should see it, and write a summary. That’s 15 to 30 minutes per inquiry. For a firm handling 40 new inquiries a week, that’s ten to twenty hours of paralegal or associate time that never gets billed.

These aren’t edge cases. They’re the baseline operating cost of running a modern law firm. The question is whether you can automate enough of this work to recover 40% to 60% of those hours without hiring more people or adding overhead.

What OpenAI Presence Actually Does

Presence is OpenAI’s first managed-service offering for enterprise AI agents. The model is simple: you describe the task you want automated, OpenAI assigns a technical team to build the agent, and that team stays embedded to maintain and improve the system over time.

The agent itself runs on OpenAI’s infrastructure. It uses their latest models, their voice API for phone-based agents, and their function-calling tools to integrate with your existing software. The technical team handles the configuration, the prompt engineering, the API connections, and the monitoring. You don’t need to hire a data scientist or train your staff on machine learning.

For law firms, the most immediate use cases are client intake, matter triage, and document review. These are tasks where the input is predictable, the decision tree is clear, and the output can be checked by a human before it goes to the client or into a case file.

An Intake Voice Agent answers every call to your main line. It greets the caller, asks a series of qualifying questions, checks for conflicts against your case management system, captures the matter details, and books a consultation directly into the partner’s calendar. The entire interaction takes three to five minutes. The caller gets a confirmation email. The partner gets a one-paragraph brief with the caller’s name, the matter type, the urgency, and the scheduled time.

One litigation firm in our network deployed a voice agent for after-hours intake and saw their conversion rate on evening and weekend calls jump from 28% to 61% in the first 90 days. The agent answered every call, captured the details, and booked the consultation while the caller was still motivated. By the time a competitor called them back the next morning, the consultation was already on the calendar.

A Matter Triage Agent reviews incoming emails and web-form submissions. It reads the inquiry, classifies the practice area, scores the fit based on case size and complexity, flags conflicts, and routes the matter to the right partner with a summary attached. The partner sees the inquiry within minutes, not hours, and can decide whether to respond immediately or delegate to an associate.

A Document Review Agent performs first-pass review on contracts, discovery documents, and matter files. It reads the document, flags key clauses, summarises positions, identifies risks, and produces a memo that an associate would typically spend four to six hours writing. The associate reviews the memo, makes edits, and delivers the final version to the partner. The client still gets associate-level work, but the firm’s cost drops by 60% to 70%.

The technical team from OpenAI configures each agent to your practice. They connect it to your case management system, your calendar, your document storage, and your CRM. They write the prompts that guide the agent’s responses. They test the system with real inquiries and adjust the logic based on how your firm actually works. Once the agent is live, they monitor performance, tune the prompts, and push updates as OpenAI releases new models.

This is the part that makes Presence different from a SaaS product. You’re not buying software and figuring it out yourself. You’re buying the engineering team that builds and maintains the system. The cost is higher than a monthly subscription, but the risk is lower. You don’t need internal AI expertise. You don’t need to hire a consultant to integrate it. You don’t need to worry about model updates breaking your workflows.

For a detailed breakdown of how voice agents handle intake workflows step-by-step, we’ve built a practical worksheet that walks through the decision tree, the data capture, and the handoff to your calendar. You can grab the AI Client Intake Checklist for Law Firms and use it to map your current intake process against what an agent can automate.

How This Compares to Building It Yourself

The alternative to Presence is building the system in-house or hiring a consultant to do it for you. Both options are viable if you have the budget and the timeline, but the economics are tricky.

Building in-house means hiring a data engineer or an AI specialist. Salary for someone with the right skills runs $120K to $180K in most markets, plus benefits and onboarding time. That person needs three to six months to build the first agent, integrate it with your systems, and get it stable enough to handle real client interactions. If they leave, you’re back to square one.

Hiring a consultant gets you faster results, but the cost is front-loaded. A good AI consultancy will charge $80K to $150K for a custom voice agent with integrations. That includes discovery, design, build, and handoff. Maintenance and updates are usually billed separately, at $5K to $15K per month depending on complexity.

Presence sits somewhere in between. OpenAI hasn’t published pricing yet, but the model is likely to be a setup fee plus a monthly retainer. The setup fee covers the initial build and integration. The retainer covers ongoing maintenance, model updates, and access to the engineering team. Based on how other managed AI services are priced, expect $40K to $80K for setup and $8K to $20K per month for the retainer.

That’s not cheap, but it’s predictable. You don’t have payroll risk. You don’t have the uncertainty of a one-time build that might need a rebuild in 18 months when the models change. You get a team that’s already expert in the technology, already familiar with the edge cases, and already plugged into OpenAI’s roadmap.

The break-even math depends on how much time you’re currently spending on the tasks the agent will automate. If you’re losing six hours a week of partner time to intake and triage, that’s $15K to $25K a month in opportunity cost at typical billing rates. If the agent recovers half of that time, you’re break-even in month two or three. If it recovers 70%, you’re profitable from day one.

For firms that want to explore what an agent could automate in their specific practice, we run a 60-minute Omni Audit that maps your intake, triage, and document workflows, identifies the highest-value automation targets, and produces a one-page implementation roadmap. You can book a 60-min Omni Audit and walk away with a clear picture of what’s possible and what it would cost.

What You Actually Get From a Managed Agent

The value of a managed service isn’t just the technology. It’s the reduction in decision fatigue and the elimination of technical risk. You don’t need to become an AI expert. You don’t need to evaluate ten different voice platforms. You don’t need to worry about prompt injection attacks or model drift or API rate limits. Someone else handles all of that.

For a law firm, this matters because your competitive advantage isn’t AI. It’s legal expertise, client relationships, and case outcomes. AI is a tool to protect your margin and scale your capacity. The less time you spend managing the tool, the more time you spend on the work that actually differentiates your practice.

A managed agent also gives you access to OpenAI’s latest models as soon as they’re released. When GPT-5 or GPT-6 comes out, your agent gets upgraded automatically. You don’t need to rewrite prompts or retrain the system. The engineering team handles the migration, tests the new model, and pushes the update. You get the performance improvement without the disruption.

The other benefit is reliability. A managed service comes with uptime guarantees, monitoring, and support. If the agent stops working at 9pm on a Friday, someone from OpenAI is on call to fix it. If a caller reports a bad experience, the team reviews the transcript, identifies the issue, and adjusts the prompt. You’re not troubleshooting API errors or debugging integration failures. You’re reviewing performance reports and deciding whether to expand the agent’s scope.

This is the model we use at Enterprise DNA for Omni, our AI agent platform for mid-market businesses. We build voice agents, ops agents, and app agents for clients, and we stay embedded to maintain and improve them over time. The client gets the automation, the cost savings, and the capacity increase. We handle the technical complexity, the integration work, and the ongoing tuning. It’s the same managed-service model OpenAI is now offering at enterprise scale.

If you want to see how this applies to law firms specifically, the AI audit for law firms walks through the intake, triage, and document workflows we typically automate, the time savings we measure, and the implementation timeline. It’s a 60-minute working session, not a sales pitch. You get three outputs: a process map, a savings estimate, and a one-page roadmap.

The Real Question Is What You Automate First

The hardest part of deploying AI agents isn’t the technology. It’s deciding which task to automate first. Most firms have a dozen candidates: intake, triage, document review, contract analysis, discovery prep, client follow-up, billing summaries, conflict checks. All of them are valid. All of them would save time. But you can’t automate everything at once, and the order matters.

The best first target is the task that meets three criteria. It’s high-volume, so the time savings compound quickly. It’s low-risk, so a mistake doesn’t cost you a client or a case. And it’s measurable, so you can prove the ROI and build confidence for the next phase.

For most law firms, that task is after-hours intake. You’re already losing 60% to 70% of those calls. An agent that captures even half of them is pure upside. The risk is low because the agent is booking consultations, not giving legal advice. And the metric is simple: calls answered, consultations booked, conversion rate.

Matter triage is the second-best target. It’s high-volume, low-risk, and directly tied to partner time. Every inquiry the agent routes and summarises is 15 to 30 minutes a partner or paralegal doesn’t spend reading emails. The savings show up immediately in your calendar.

Document review is higher-value but higher-risk. The agent is producing work product that goes into a case file or gets delivered to a client. You need a human review step, and you need confidence that the agent’s output is reliable. That makes it a better second or third phase, after you’ve proven the model with intake and triage.

The firms that get the most value from AI agents are the ones that start narrow, measure everything, and expand based on results. They don’t try to automate the entire practice in month one. They pick one workflow, deploy the agent, track the time savings and the error rate, and then move to the next target. By month six, they’ve automated three or four workflows and recovered 20 to 30 hours a week of billable time.

We’ve written more about how to prioritise automation targets and build a phased rollout plan in our AI implementation guides. The short version is: start with the task that’s costing you the most time right now, prove the ROI, and expand from there.

What Happens Next

OpenAI Presence is live now for enterprise customers. If you’re running a law firm doing $1M to $25M in revenue, you’re in the target range. The service is designed for businesses that have clear automation targets, predictable workflows, and the budget to pay for a managed service instead of a DIY product.

The first step is deciding whether this model makes sense for your practice. That depends on how much time you’re currently losing to intake, triage, and document prep, and whether you have the internal capacity to manage a build-it-yourself project. If you’re losing six hours a week of partner time and you don’t have a data engineer on staff, a managed service is probably the faster path to ROI.

The second step is mapping your workflows and identifying the highest-value automation targets. This is where most firms get stuck. You know the work is inefficient, but you don’t have a clear picture of where the time goes or which tasks are automatable. That’s what the Omni Audit is for. We spend 60 minutes mapping your intake, triage, and document workflows, we identify the tasks an agent can handle, and we produce a one-page roadmap with time savings and cost estimates.

You can book my Omni Audit and walk away with a clear picture of what’s possible for your practice. No deck, no pitch, just three outputs: a process map, a savings estimate, and a roadmap. If you decide to move forward with Presence or another platform, you’ll have the clarity you need to make the right call. If you decide to wait, you’ll have a benchmark for when the economics make sense.

The broader point is that managed AI services are now a real option for mid-market businesses. You don’t need to hire a data team. You don’t need to become an AI expert. You describe the work you want automated, and someone else builds and maintains the system. For law firms losing $80K to $250K a year to unbilled administrative time, that’s a model worth evaluating.

If you want to explore what this looks like for your practice, see Omni for law firms and book a working session. We’ll map the work, estimate the savings, and give you a roadmap you can use whether you build with us, with OpenAI, or with another provider. The goal is clarity, not a sales cycle.

For more on how AI agents are reshaping professional services, check out the latest updates on our insights page or dive into the technical details of Omni Voice and Omni Ops. The technology is ready. The question is whether your practice is ready to deploy it.